Merge pull request #524 from seek4science/rocrate-jsonld

Rocrate jsonld

app/models/concerns/workflow_extraction.rb

@@ -76,7 +76,7 @@ def populate_ro_crate(crate)
     wf = crate.main_workflow || ROCrate::Workflow.new(crate, c.filepath, c.original_filename)
     wf.content_size = c.file_size
     crate.main_workflow = wf
-    crate.main_workflow.programming_language = ROCrate::ContextualEntity.new(crate, nil, workflow_class&.ro_crate_metadata || Seek::WorkflowExtractors::Base::NULL_CLASS_METADATA)
+#    crate.main_workflow.programming_language = ROCrate::ContextualEntity.new(crate, nil, extractor_class.ro_crate_metadata)
 
     begin
       d = diagram
@@ -88,22 +88,55 @@ def populate_ro_crate(crate)
     rescue WorkflowDiagram::UnsupportedFormat
     end
 
-    authors = creators.map { |person| crate.add_person(nil, person.ro_crate_metadata) }
-    others = other_creators&.split(',')&.collect(&:strip)&.compact || []
-    authors += others.map.with_index { |name, i| crate.add_person("creator-#{i + 1}", name: name) }
-    crate.author = authors
-    crate['provider'] = projects.map { |project| crate.add_organization(nil, project.ro_crate_metadata).reference }
-    crate.license = license
     crate.identifier = ro_crate_identifier
     crate.url = ro_crate_url('ro_crate')
-    crate['isBasedOn'] = source_link_url if source_link_url
-    crate['sdPublisher'] = crate.add_person(nil, contributor.ro_crate_metadata).reference
-    crate['sdDatePublished'] = Time.now
+
+    workflow = is_a_version? ? self.parent : self
+    merge_entities(crate, workflow) if workflow
+
+    crate['isBasedOn'] = source_link_url if source_link_url && !crate['isBasedOn']
+    crate['sdDatePublished'] = Time.now unless crate['sdDatePublished']
     crate['creativeWorkStatus'] = I18n.t("maturity_level.#{maturity_level}") if maturity_level
 
-    crate.preview.template = PREVIEW_TEMPLATE
+    # brute force deletion as I cannot track down where it comes from
+    crate.contextual_entities.delete_if { |c| c['@id'] == '#ro-crate-preview.html' }
+    crate
   end
 
+  def merge_entities(crate, workflow)
+    workflow_struct = Seek::BioSchema::Serializer.new(workflow).json_representation
+
+    context = {
+      '@vocab' => 'https://somewhere.com/'
+    }
+    workflow_struct['@context'] = context
+    crate['name'] = "Research Object Crate for #{workflow_struct['name']}"
+    crate['description'] = workflow_struct['description']
+
+    workflow_struct.except!('encodingFormat')
+
+    flattened = JSON::LD::API.flatten(workflow_struct, context)
+    flattened.except!('@context')
+
+    flattened['@graph'].each do |elem|
+      type = elem['@type']
+      type = [type] unless type.is_a?(Array)
+      if type.include?('ComputationalWorkflow')
+        merge_fields(crate.main_workflow, elem)
+      else
+        entity_class = ROCrate::ContextualEntity.specialize(elem)
+        entity = entity_class.new(crate, elem['@id'], elem)
+        crate.add_contextual_entity(entity)
+      end
+    end
+  end
+
+  def merge_fields(crate_workflow, bioschemas_workflow)
+    bioschemas_workflow.each do |key, value|
+      crate_workflow[key] = value unless crate_workflow[key]
+    end
+  end
+
   def ro_crate
     inner = proc do |crate|
       populate_ro_crate(crate) if should_generate_crate?

app/models/workflow_crate_builder.rb

@@ -20,7 +20,7 @@ def build
       crate.main_workflow = ROCrate::Workflow.new(crate, workflow[:data], get_unique_filename(crate, workflow))
       crate.main_workflow.programming_language = crate.add_contextual_entity(ROCrate::ContextualEntity.new(crate, nil, workflow_class&.ro_crate_metadata || Seek::WorkflowExtractors::Base::NULL_CLASS_METADATA))
       crate.main_workflow['url'] = workflow[:data_url] if workflow[:data_url].present?
-      if diagram[:data].present?
+      if diagram && diagram[:data].present?
         crate.main_workflow.diagram = ROCrate::WorkflowDiagram.new(crate, diagram[:data], get_unique_filename(crate, diagram))
         crate.main_workflow.diagram['url'] = diagram[:data_url] if diagram[:data_url].present?
       end

lib/seek/bio_schema/data_catalog_mock_model.rb

@@ -22,7 +22,8 @@ def provider
         {
           '@type' => 'Organization',
           'name' => Seek::Config.dm_project_name,
-          'url' => Seek::Config.dm_project_link
+          'url' => Seek::Config.dm_project_link,
+          '@id' => Seek::Config.dm_project_link
         }
       end
 

lib/seek/bio_schema/resource_decorators/base_decorator.rb

@@ -5,6 +5,7 @@ module ResourceDecorators
       # The Decorator is an extension to the resource that provided or alters the properties of that resource
       # for Schema.org (Bioschemas.org)
       class BaseDecorator
+
         include ActionView::Helpers::SanitizeHelper
         include Rails.application.routes.url_helpers
 
@@ -26,7 +27,7 @@ def attributes
 
         # The @context to be used for the JSON-LD
         def context
-          'http://schema.org'
+          Seek::BioSchema::Serializer::SCHEMA_ORG
         end
 
         # The schema.org @type .

lib/seek/bio_schema/resource_decorators/creative_work.rb

@@ -3,6 +3,7 @@ module BioSchema
     module ResourceDecorators
       # Decorator that provides extensions for a Event
       class CreativeWork < Thing
+
         associated_items producer: :projects
 
         schema_mappings license: :license,
@@ -11,7 +12,8 @@ class CreativeWork < Thing
                         created_at: :dateCreated,
                         updated_at: :dateModified,
                         content_type: :encodingFormat,
-                        subject_of: :subjectOf
+                        subject_of: :subjectOf,
+                        provider: :sdPublisher
 
         def content_type
           return unless resource.respond_to?(:content_blob) && resource.content_blob
@@ -25,7 +27,7 @@ def license
 
         def all_creators
           others = other_creators&.split(',')&.collect(&:strip)&.compact || []
-          others = others.collect { |name| { "@type": 'Person', "name": name } }
+          others = others.collect { |name| { "@type": 'Person', "@id": "##{ROCrate::Entity.format_id(name)}", "name": name } }
           all = (mini_definitions(creators) || []) + others
           return if all.empty?
           all

lib/seek/bio_schema/resource_decorators/event.rb

@@ -3,15 +3,23 @@ module BioSchema
     module ResourceDecorators
       # Decorator that provides extensions for a Event
       class Event < Thing
+
+        EVENT_PROFILE = 'https://bioschemas.org/profiles/Event/0.2-DRAFT-2019_06_14/'
+
         associated_items contact: :contributors,
                          host_institution: :projects
         schema_mappings contact: :contact,
                         start_date: :startDate,
                         end_date: :endDate,
                         event_type: :eventType,
                         location: :location,
-                        host_institution: :hostInstitution
+                        host_institution: :hostInstitution,
+                        conformsTo: "dct:conformsTo"
 
+        def conformsTo
+          EVENT_PROFILE
+        end
+
         def contributors
           [contributor]
         end

lib/seek/bio_schema/resource_decorators/person.rb

@@ -3,14 +3,21 @@ module BioSchema
     module ResourceDecorators
       # Decorator that provides extensions for a Person
       class Person < Thing
+        PERSON_PROFILE = 'https://bioschemas.org/profiles/Person/0.2-DRAFT-2019_07_19/'
+
         associated_items member_of: :projects
 
         schema_mappings first_name: :givenName,
                         last_name: :familyName,
                         image: :image,
                         member_of: :memberOf,
-                        orcid: :orcid
+                        orcid: :orcid,
+                        conformsTo: "dct:conformsTo"
 
+        def conformsTo
+          PERSON_PROFILE
+        end
+
         def url
           web_page.blank? ? identifier : web_page
         end

lib/seek/bio_schema/resource_decorators/thing.rb

@@ -30,6 +30,15 @@ def keywords
           tags_as_text_array.join(', ') if resource.respond_to?(:tags_as_text_array)
         end
 
+        def provider
+          {
+            '@type' => 'Organization',
+            '@id' => Seek::Config.site_base_host,
+            'name' => Seek::Config.dm_project_name,
+            'url' => Seek::Config.site_base_host
+          }
+        end
+
         def date_created
           resource.date_created.try(:utc) if resource.respond_to?(:date_created)
         end

lib/seek/bio_schema/resource_decorators/workflow.rb

@@ -3,47 +3,67 @@ module BioSchema
     module ResourceDecorators
       # Decorator that provides extensions for a Workflow
       class Workflow < CreativeWork
-        associated_items sd_publisher: :contributors
 
-        schema_mappings sd_publisher: :sdPublisher,
-                        version: :version,
+        WORKFLOW_PROFILE = 'https://bioschemas.org/profiles/ComputationalWorkflow/1.0-RELEASE/'
+
+        FORMALPARAMETER_PROFILE = 'https://bioschemas.org/profiles/FormalParameter/1.0-RELEASE/'
+
+        schema_mappings version: :version,
                         image: :image,
                         programming_language: :programmingLanguage,
+                        producer: :producer,
                         inputs: :input,
-                        outputs: :output
+                        outputs: :output,
+                        license: :license,
+                        conformsTo: "dct:conformsTo"
 
         def contributors
           [contributor]
         end
 
+        def conformsTo
+          WORKFLOW_PROFILE
+        end
+
         def image
           return unless resource.diagram_exists?
           diagram_workflow_url(resource, version: resource.version, host: Seek::Config.site_base_host)
         end
 
         def schema_type
-          'ComputationalWorkflow'
+          ['File', 'SoftwareSourceCode', 'ComputationalWorkflow']
         end
 
         def programming_language
-          resource.workflow_class&.title
+          resource.workflow_class&.extractor_class.ro_crate_metadata
         end
 
         def inputs
-          formal_parameters(resource.inputs)
+          formal_parameters(resource.inputs, 'inputs')
         end
 
         def outputs
-          formal_parameters(resource.outputs)
+          formal_parameters(resource.outputs, 'outputs')
+        end
+
+        def license
+          Seek::License.find(resource.license)&.url
         end
 
         private
 
-        def formal_parameters(properties)
+        def formal_parameters(properties, group_name)
+          if self.title
+            wf_name = self.title.downcase.gsub(/[^0-9a-z]/i, '_')
+          else
+            wf_name = 'dummy'
+          end
           properties.collect do |property|
             {
               "@type": 'FormalParameter',
-              name: property.id
+              "@id": "##{wf_name}-#{group_name}-#{property.id}",
+              name: property.name || property.id,
+              "dct:conformsTo": FORMALPARAMETER_PROFILE
             }
           end
         end

lib/seek/bio_schema/serializer.rb

@@ -7,22 +7,28 @@ class Serializer
       include ActionView::Helpers::SanitizeHelper
       attr_reader :resource
 
+      SCHEMA_ORG = 'https://schema.org'
+
       # initialise with a resource
       def initialize(resource)
         @resource = resource
       end
 
+      def json_representation
+        repr = {
+          '@context' => resource_decorator.context,
+          '@type' => resource_decorator.schema_type
+        }.merge(attributes_json)
+        repr.deep_stringify_keys
+      end
+
       # returns the JSON-LD as a String, for the resource
       def json_ld
         unless supported?
           raise UnsupportedTypeException, "Bioschema not supported for #{resource.class.name}"
         end
-        json = {
-          '@context' => resource_decorator.context,
-          '@type' => resource_decorator.schema_type
-        }.merge(attributes_json)
 
-        JSON.pretty_generate(json)
+        JSON.pretty_generate(json_representation)
       end
 
       # whether the resource BioSchema was initialized with is supported

test/functional/workflows_controller_test.rb

@@ -419,7 +419,7 @@ def setup
       post :create_ro_crate, params: {
           ro_crate: {
               workflow: { data: fixture_file_upload('files/checksums.txt') },
-              diagram: { data: fixture_file_upload('files/file_picture.png') },
+                          diagram: { data: fixture_file_upload('files/file_picture.png') },
               abstract_cwl: { data: fixture_file_upload('files/workflows/rp2-to-rp2path-packed.cwl') }
           },
           workflow_class_id: cwl.id
@@ -584,7 +584,8 @@ def setup
     end
     assert_response :success
     assert wf = assigns(:workflow)
-    crate_workflow = wf.ro_crate.main_workflow
+    workflow_crate = ROCrate::WorkflowCrateReader.read_zip(wf.content_blob.path)
+    crate_workflow = workflow_crate.main_workflow
     assert crate_workflow
     assert_equal 'file%20with%20spaces%20in%20name.txt', crate_workflow.id
   end
@@ -639,16 +640,16 @@ def setup
       post :create_ro_crate, params: {
           ro_crate: {
               workflow: { data: fixture_file_upload('files/workflows/rp2-to-rp2path-packed.cwl') },
-              diagram: { data: fixture_file_upload('files/file_picture.png') },
+                          diagram: { data: fixture_file_upload('files/file_picture.png') },
               abstract_cwl: { data: fixture_file_upload('files/workflows/rp2-to-rp2path-packed.cwl') }
           },
           workflow_class_id: cwl.id
       }
     end
     assert_response :success
-    assert wf = assigns(:workflow)
-    crate_workflow = wf.ro_crate.main_workflow
-    crate_cwl = wf.ro_crate.main_workflow_cwl
+    workflow_crate = ROCrate::WorkflowCrateReader.read_zip(assigns(:workflow).content_blob.path)
+    crate_workflow = workflow_crate.main_workflow
+    crate_cwl = workflow_crate.main_workflow_cwl
     assert_not_equal crate_workflow.id, crate_cwl.id
   end
 
@@ -760,15 +761,16 @@ def setup
       post :create_ro_crate, params: {
           ro_crate: {
               workflow: { data_url: 'https://github.com/bob/workflow/blob/master/workflow.txt' },
-              diagram: { data_url: 'https://github.com/bob/workflow/blob/master/diagram.png' },
+                          diagram: { data_url: 'https://github.com/bob/workflow/blob/master/diagram.png' },
               abstract_cwl: { data_url: 'https://github.com/bob/workflow/blob/master/abstract.cwl' }
           },
           workflow_class_id: cwl.id
       }
     end
     assert_response :success
     assert wf = assigns(:workflow)
-    crate_workflow = wf.ro_crate.main_workflow
+    workflow_crate = ROCrate::WorkflowCrateReader.read_zip(wf.content_blob.path)
+    crate_workflow = workflow_crate.main_workflow
     assert crate_workflow
     assert_equal 'workflow.txt', crate_workflow.id
   end

test/unit/bio_schema/data_catalog_mock_model_test.rb

@@ -32,9 +32,10 @@ def setup
     with_config_value(:dm_project_name, 'WIBBLE') do
       with_config_value(:dm_project_link, 'http://wibble.eu') do
         expected = {
-          '@type' => 'Organization',
-          'name' => 'WIBBLE',
-          'url' => 'http://wibble.eu'
+          "@type"=>"Organization",
+          "@id"=>"http://wibble.eu",
+          "name"=>"WIBBLE",
+          "url"=>"http://wibble.eu"
         }
         assert_equal expected, @data_catalogue.provider
       end